MINERVAAIPublic healthcare data is effectively free, and every tool built on it will answer any question you ask — whether or not the underlying file supports an answer. The failure mode is not a missing number. It is a plausible one, produced with no indication that it should not be trusted.
Every engagement adds dated, sourced determinations about where the public record misleads. They are the reason the answers hold, and they compound. A sample from the ledger:
Prevents — a rate comparison presented to a board as competitive intelligence when it is an artifact of rate setting.
Prevents — understating share on the fastest-growing half of Medicare, where the free federal file now has broader coverage than several paid products.
Prevents — counsel citing a withdrawn requirement in correspondence with a state attorney general.
Prevents — commissioning a hand-built corpus that can be assembled from an index already in the public record.
Most tools answer the questions with one obvious source. The questions that decide capital allocation require two sources joined on a spine nobody maintains. Maintaining it is the work.
Two questions we do not answer: physician-level referral flow and all-payer volume. Both require licensed claims. You will hear that in the first meeting rather than the fourth.
How the platform worksWhy assembly of the public record is no longer a differentiator, and why the determinations layer — what the data cannot support — is what separates usable analysis from plausible error.
Penalty scores and payment adjustment factors live in different files. Multiplying them is the difference between a rating and a budget line.
Price files are overwritten monthly and most vendors keep one snapshot. History is the asset that can only be started.
Sixty minutes. Bring the question you could not get a straight answer to last quarter, and we will work it live — with the sources, the dates, and a clear statement of what we will not claim.